By Roland Brandli, Global Strategy Director, Smartstream
At the MEA Finance Banking Technology Summit in Dubai this week, I sat on a panel where the conversation kept returning to the same point. Payments is not under one kind of pressure. It is under four, all at once. And the way institutions respond to each will determine where their operations stand in five years.
Competition has moved into new territory
Banks have always competed on payments. Cross-border flows, domestic clearing, regional schemes: these are familiar battlegrounds. What has changed is who is showing up. Scheme providers including Mastercard and Visa are pushing into spaces they previously left to banks. Meanwhile, multinational corporates are asking a different question entirely: why use the correspondent banking chain at all?
Stablecoins and programmable payments make that question real. Moving money globally within an organisation, or across a client base, without touching traditional rails is no longer a concept being debated in white papers. It is being piloted. In some cases, it is already running.
Where it all lands is genuinely uncertain. The regulatory boundaries around stablecoins are coming into view. What the flow of money inside those boundaries looks like, and where banks fit in, is still open. Institutions that wait for full clarity before preparing their operations will arrive too late.
Cost is higher than most operations budgets reflect
A modern payments operation runs multiple platforms in parallel: cross-border systems, local and regional clearing, instant payments rails. Each carries its own integration requirements, its own compliance stack — KYC, AML, pre-authorisation, liquidity controls. The cumulative cost of keeping all of it current is significant and growing.
The cost that tends to be underestimated, though, is exception management. Payments that fail, flag, or fall outside automated processing flows still get resolved largely by people. They chase counterparties, work through unstructured messages, and reconcile outcomes by hand. In many operations, a substantial portion of headcount sits in this work. So does a substantial portion of customer satisfaction risk.
Smart Agents are making a measurable difference here. We are seeing exception management costs fall by 50 to 70 per cent, with AI taking on the manual workload so that operations teams can focus on decisions that actually require judgement.
Time used to be a resource. Now it is a constraint.
For most of banking history, settlement cycles gave operations teams room to work. Errors could be caught and corrected. Liquidity could be managed. Exceptions could be investigated before anything became urgent.
Instant payments removed that room. When settlement is measured in seconds, the window to intervene closes before most manual processes have even started.
Addressing this requires more than speed. It requires removing the manual steps from exception resolution entirely, so that when something needs a human decision, the person is brought in at the moment of choice rather than at the start of a lengthy investigation. The goal is not to remove people from payments operations. It is to put them where they can add something that a machine cannot.
Complexity is compounding faster than expertise can keep up
Every new payment rail an institution adds brings integration work, maintenance overhead, and a need for people who understand how that rail behaves: technically, operationally, and in terms of how it fails. The problem is that rails are multiplying faster than the talent pool that supports them.
Part of the answer is capturing institutional knowledge in systems rather than leaving it in people. But architecture matters just as much, and this is where the AI conversation gets complicated.
Payments carry critical customer data. If you want to deploy AI agents across your exception management workflows, you face a real challenge: how do you use AI without that data leaving your environment and going to a cloud-based model? It can be solved. But it requires deliberate choices and sustained attention. Institutions that treat data governance as an afterthought will either find themselves exposed or unable to realise the operational benefits they are chasing.
How Smartstream supports payments operations
These four pressures do not resolve on their own. But they share a common thread: the need for control across every payment rail, without operational overhead growing in step with volume.
Smart Payments gives operations teams a single point of control across multiple rails, with automated exception management and real-time visibility.
